OBJECTIVETo study the factors of colorectal cancer (CRC) after radical resection to provide data predicting the prognosis of the patients.

METHODS120 cases of CRC were collected in this study. Medical clinical records and 5-year follow-up data were reviewed. Streptavidin-peroxidase immunohistochemical technique was used to detect the expression of p53, C-erbB-2, nm23-H(1) and Ras on formalin-fixed, paraffin embedded sections of CRC from the 120 patients.

RESULTSResults showed that the rates of positive expression of p53, C-erbB-2, Ras and nm23-H(1) were 62.5% (75/120), 25.8% (31/120), 80.0% (96/120) and 60.8% (73/120) respectively in the CRC tissue. All pathological variables and biological markers were analyzed with Cox regression model (alpha = 0.05). Eight distinguished prognostic factors were identified in the univariate analysis as: macroscopic configuration, histology grade, depth of invasion of intestinal, lymph nodes metastasis, Dukes' classification, p53, Ras and nm23-H(1). The results of multivariate analysis (alpha = 0.05) indicated that the independent prognostic factors were Dukes' classification, p53 and nm23-H(1) (P = 0.000), with relative risk of 3.06, 6.02 and 0.40, respectively. A prognostic model: h(t, x) = h(0)(t)exp (-0.9269X(14) + 1.1197X(10) + 1.7948X(11)) was established. Sensitivity, specificity agreement and reliability of the model and Kappa were 79.1%, 83.0%, 80.8% and 0.62, respectively.

CONCLUSIONDukes' classification, p53 and nm23-H(1)seemed to be independent and important prognostic factors. This prognostic model could be used to evaluate the prognosis of patients with CRC by clinicians.